22. September 2012
Begin with the End in Mind
I am often asked “How do you get Ruby to do that”. The short answer is I use positive reinforcement, a clicker, treats, love and patience. Here is a short video, shot in real time, for how we trained Ruby on one behavior. It builds on skills that we had previously taught her and serves as a good example of how effective clicker training can be.
Steven Covey popularized the concept of beginning with the end in mind. The idea is to visualize your successful outcome. I often use this in dog training. The other way to think about how to train a behavior is to imagine the progression to the desired outcome as a flip-book. Drawn on the last page is the picture of what you want the dog to do. Drawn on the first is where the dog starts. It is up to you to draw the intermediate pictures and train your dog for each step along the way. You may chose to start at the last behavior and work backwards. This is called “back chaining” and is commonly used to train retrieve. Teach the dog to give you the ball first, then work backwards to the throw.
In this video Mary Jane and I teach Ruby to go forward, further and further to a target hidden under the peninsula. I start with the target in hand and reinforce the right behavior with a clicker. Ruby knows the click means she did what I wanted and a treat is on the way. It marks the exact right behavior. (The same technique is used for Killer Whales, you’ve heard the “peep” of the whistle.)
We taught Ruby a “Target” command as part of her agility training to show her where I wanted her to go (or stop). It is a very useful command for videography too. Using a hidden target, I can send her to a spot in the scene. The target itself does not matter. I usually use a can lid, but I have also used a laser pointer, a piece of tape, anything that she can differentiate in the environment.
In our kitchen we installed under-counter lights that are triggered by a touch dimmer under the peninsula. When I get up in the morning it’s dark and the dimmer is an easy mark for me to hit. I figured it would be easy for Ruby too.
There are many, many resources for clicker training, I don’t intend this to be all inclusive instruction. You should:
- Keep the sessions short. In my case I keep it to 10 minutes or less
- Use plenty of small rewards
- Keep it fun
- End on a success
Mary Jane captured some cool things on the video if you watch Ruby. There are a few moments where she stops and thinks, then acts. I love to watch the wheels turn. Of course I love the sudden bursts of exuberance, the pounce, and the kisses!
With the exception of the part where I cut the lid, I left the video in real time so you can see the training progression. This is a short training, less than 7 minutes. Following this session, we did very short reinforcements each day. (Three commands at most, less than a minute.) Mary Jane added “Light” once Ruby was performing the behavior consistently, after about 2 days. In the mornings, in the dark, I would send Ruby to the target and reward success. At the end of a week we had a consistent “Get the Lights” command, and I don’t run the risk of stubbing my toes on the way to make coffee anymore.